Tests in Australia on the effectiveness of a screening questionnaire have highlighted the difficulties parents and teachers alike face when they try to predict which children are most prone to problems.

Prevention is invariably complicated – and it can be costly. Either you offer a program to all the children in a school (or other group) in the hope of benefiting some of them; or, to save money, you provide the program only to the ones who appear to be most at risk.

But detecting which children are most at risk brings its own difficulties. If the prediction method is awry, you run the twin dangers of wasting resources on and causing distress to children who are mislabeled as needy, and of not helping children those wrongly identified as OK. Unfortunately, far too little is known about the accuracy of many screening methods.

Sarah Dwyer and her colleagues School of Public Health, Queensland University of Technology attempted to remedy this situation by studying the accuracy of one such device: the Family Risk Factor Checklist (FRFC).

FRFC asks teachers and parents 48 questions about a child to assess whether he/she is at risk for mental health problems. They compared FRFC to another method: asking parents and teachers just one question: "Do you think that this child has a higher chance than average of developing a behavioral, emotional, or mental health problem in the future?" The study involved 766 children, aged between four and eight, living near Brisbane.

Reported in Prevention Science, what they found should be borne in mind by anyone considering applying a screening questionnaire such as FRFC. Neither parents nor teachers were very good at identifying children who developed depression, anxiety, and other 'internal disorders' a year later. They were somewhat better at predicting behavior problems, correctly identifying about 70 percent of children who later had some such experience.

But, interestingly, parents were no better at predicting whether their children would develop behavior problems, if they were asked 48 questions or just the one. Moreover, teachers did a better job predicting which kids would have behavior problems a year hence when they answered just the one general question.

The researchers suggest that, before turning to a screening questionnaire, institutions should consider how much accuracy they want and whether one question might be as good as many.